Article

pubs.acs.org/est

Tradeoffs and Synergies between Biofuel Production and Large Solar

Infrastructure in Deserts
Sujith Ravi,*,† David B. Lobell,†,‡ and Christopher B. Field†,§
†
Department of Environmental Earth System Science, Stanford University, 473 Via Ortega, Stanford, California 94305, United States
‡
Center on Food Security and the Environment, Stanford University, Stanford, California 94305, United States
§
Department of Global Ecology, Carnegie Institution for Science, Stanford, California 94301, United States
*
S Supporting Information

ABSTRACT: Solar energy installations in deserts are on the rise, fueled by

technological advances and policy changes. Deserts, with a combination of high
solar radiation and availability of large areas unusable for crop production are ideal
locations for large solar installations. However, for efficient power generation,
solar infrastructures use large amounts of water for construction and operation.
We investigated the water use and greenhouse gas (GHG) emissions associated
with solar installations in North American deserts in comparison to agave-based
biofuel production, another widely promoted potential energy source from arid
systems. We determined the uncertainty in our analysis by a Monte Carlo
approach that varied the most important parameters, as determined by sensitivity
analysis. We considered the uncertainty in our estimates as a result of variations in
the number of solar modules ha−1, module efficiency, number of agave plants
ha−1, and overall sugar conversion efficiency for agave. Further, we considered the
uncertainty in revenue and returns as a result of variations in the wholesale price
of electricity and installation cost of solar photovoltaic (PV), wholesale price of agave ethanol, and cost of agave cultivation and
ethanol processing. The life-cycle analyses show that energy outputs and GHG offsets from solar PV systems, mean energy
output of 2405 GJ ha−1 year−1 (5 and 95% quantile values of 1940−2920) and mean GHG offsets of 464 Mg of CO2 equiv ha−1
year−1 (375−562), are much larger than agave, mean energy output from 206 (171−243) to 61 (50−71) GJ ha−1 year−1 and
mean GHG offsets from 18 (14−22) to 4.6 (3.7−5.5) Mg of CO2 equiv ha−1 year−1, depending upon the yield scenario of agave.
Importantly though, water inputs for cleaning solar panels and dust suppression are similar to amounts required for annual agave
growth, suggesting the possibility of integrating the two systems to maximize the efficiency of land and water use to produce both
electricity and liquid fuel. A life-cycle analysis of a hypothetical colocation indicated higher returns per m3 of water used than
either system alone. Water requirements for energy production were 0.22 L MJ−1 (0.28−0.19) and 0.42 L MJ−1 (0.52−0.35) for
solar PV−agave (baseline yield) and solar PV−agave (high yield), respectively. Even though colocation may not be practical in all
locations, in some water-limited areas, colocated solar PV−agave systems may provide attractive economic incentives in addition
to efficient land and water use.

■ INTRODUCTION
Energy production using fossil fuels is a major contributor to
impact land and water resources.6−8 Moreover, renewable
energy technologies differ considerably in their effectiveness in
anthropogenic greenhouse gas (GHG) emissions and asso- specific geographic locations and in their environmental and
ciated global warming.1 In this regard, low carbon emission socioeconomic impacts. Thus, to provide all of our energy
technologies, such as biofuels and solar energy, may provide needs, it is necessary to approach renewable energy develop-
alternative pathways for sustainable energy production to meet ment through a combination of complementary technologies to
the current and future energy requirements.1 Solar and biofuel maximize returns on resource use and to minimize environ-
technologies both use the energy from the sun, using mental impacts.1,2
photovoltaic (PV) conversion to electricity in the case of Solar energy installations in drylands are on the rise, fueled
solar energy and harvesting plant biomass that can be processed by technological advances and policy changes.9 Drylands, with a
as solid or liquid fuels in the case of biofuel energy.2 Even combination of high solar radiation and availability of large
though renewable energy potentially provides several positive areas unusable for crop production, are ideal locations for large
aspects, such as reduction of GHGs, reclamation of degraded solar installations. Drylands cover over 40% of the earth’s land
land, increased energy independence, employment opportu-
nities, acceleration of rural electrification, and improvement of Received: March 30, 2013
the quality of life in developing countries,3−5 the deployment of Accepted: January 27, 2014
large-scale renewable energy infrastructure may negatively Published: January 27, 2014

© 2014 American Chemical Society 3021 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030
Environmental Science & Technology Article

Figure 1. Conceptual framework for analyzing the tradeoffs and synergies of colocation.

surface and support a human population of over 2 billion.10 Agave spp., perennial evergreen xerophytic plants, which have
These areas are characterized by the lowest levels of human ecological and physiological adaptations to achieve meaningful
well-being, and solar-based technologies (e.g., solar-powered yields on marginal lands, are a case in point.16,22,23 In particular,
drip irrigation and solar-powered water-pumping systems) are agave plants have high water use efficiencies as a consequence
being used in some areas as strategies for enhancing food of using the crassulacean acid metabolism (CAM) photo-
security and poverty reduction.4,11 However, solar infra- synthetic pathway, which enable them to open stomata at night
structures have a large land footprint, and for efficient power when evaporative demand is lower.21,22 Agaves are common in
generation, they use large amounts of water for construction arid and semi-arid regions of the world, including the North
and operation, mostly for cleaning solar panels or mirrors and American deserts, where they have been cultivated for fiber and
dust suppression from disturbed soils. alcoholic beverages for centuries.21,22 Even though maximum
Dust accumulation on solar panels is a major factor affecting yields are limited to areas with precipitation exceeding 1000
power output from large solar installations, and annual loss mm, reasonable productivities occur in areas with precipitation
in the range of 300−800 mm.15,16 Several characteristics of
estimates as a result of dust accumulation range from 5 to
agave make them attractive as a biofuel crop in deserts. The list
35%.12−14 While there are many strategies to minimize dust
includes significant biomass production with little or no
accumulation on PV panels, washing panels with water remains irrigation, tolerance to droughts and high temperature, ability
the main approach in most systems.14 Ibrahim et al. to make use of small precipitation events or light irrigation, and
demonstrated that washing panels every 2 months improved high sugar and cellulose content (less energy to extract sugars
efficiency but still resulted in energy output 25% below the and ethanol).16,21,22
maximum possible and, hence, recommended a cleaning Colocated solar energy and agave ethanol infrastructure
frequency of once a week for extreme arid climates.13 Thus, could maximize the efficiency of water use in drylands by
water for washing panels and dust suppression is potentially a coupling water use for cleaning panels and irrigation,
major water demand and a large component of the water minimizing dust generation by increasing soil moisture and
budget of solar facilities in desert regions,14 and it may place a vegetation cover, minimizing impacts on natural areas by
major demand on the already scarce local water resources or deploying biofuel cultivation in existing large solar infra-
may displace water allocated for domestic use or agricultural structures, and simulating economic returns to improve
activities. livelihoods in rural areas. However, to explore the logistic
Although some of the water applied to panels is evaporated, and economic feasibility of integrated solar−agave biofuel
much of it runs off the panels and into the desert soil. A systems, detailed life-cycle analyzes are needed. Here, we
pertinent question is therefore whether the moisture inputs are conduct detailed life-cycle analysis for solar PV, agave-derived
sufficient to maintain agriculture or biofuel production in water- biofuel, and a hypothetical colocated solar−agave system to
limited regions. Most solar facilities are sited in marginal lands, explore the tradeoffs and synergies (in the context of energy,
which are unusable for most crops or pasture grasses. However, water, and GHG emissions) between these two emerging land
there is a growing interest to grow biofuel feedstocks in uses (Figure 1).
marginal lands that can be cultivated without competition for
key resources for food crops.15−17 Biofuels are thought to be an
integral part of the future energy mix,18,19 and several plant
■ MATERIALS AND METHODS
Land Footprint and Water Use of Large Solar
species have been identified and cultivated to use as feedstock Installations. To estimate the land footprint and water use
for bioethanol production.15,20 Producing biomass feedstocks in large solar installations in drylands, we compiled data from
for bioethanol in drylands may be a way to partially meet the project-planning reports, from the California Energy Commis-
demand for renewable fuels without negative impacts on food sion24 and Bureau of Land Management,25 of large solar PV
production.2,15,21 and concentrated solar power (CSP) projects in the south-
3022 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030
Environmental Science & Technology Article

western United States, which are approved for construction or irrigation is provided at a rate of 100 and 550 mm precipitation
are already under construction. Our compilation covers both (annual) equivalent, respectively, in addition to background
the land area used for the solar infrastructure and the right of precipitation. In the baseline- and high-yield scenarios, stillage
way (ROW) land area allocated for additional support facilities, generated at the biorefinery stage is added as additional
such as transmission lines and roads. For the land footprint and irrigation water and fertilizer26,27 (see pages S5−S13 of the
water use, our analysis included 28 large solar installations (13 Supporting Information).
PV, 10 parabolic trough CSP, and 5 central tower CSP). ROW Ethanol production equipment and processes are assumed to
land area for 10 solar installations were used to derive a linear be comparable to those in a typical sugarcane ethanol plant.26,27
relationship between land area under the installation and ROW Agave stems and fresh leaves are used for ethanol production.
allotted to the solar companies. The annual water uses for large The bagasse generated from the extraction process and the
solar facilities were partitioned into construction and operation unused residual leaves are combusted to provide the process
phases, assuming a 30 year life for the installations. Water use in energy, and the surplus electricity is exported to the grid. The
the construction phase is mostly for dust suppression from overall sugar utilization efficiency, which includes efficiency in
disturbed soils, and water use in the operation phase is for extraction, hydrolysis, fermentation, and distillation, is assumed
cleaning panels or mirrors. The data included 5 solar PV to be 80%. This efficiency is lower than the industry average
installations and 10 CSP installations (3 CSP parabolic trough value of 90% in the case of Brazilian sugarcane ethanol
with wet cooling, 3 CSP parabolic trough with dry cooling, and production,26,27 because commercial agave-derived ethanol
4 CSP central tower) (details on solar technologies and processing pathways are not currently well-established. The
methods are on page S3 of the Supporting Information). first stage in ethanol production involves the use of diffusers to
Life-Cycle Analysis of Agave-Derived Ethanol. Fuel extract juice from uncooked agave stems and leaves. For this
ethanol production from agave is not well-established, and we study, we consider the diffuser system instead of the mill system
adopted the life-cycle analysis methodology followed by Yan et typically used in the sugarcane industry, because the diffuser
al., which is based on the production pathway used in the system yields a higher sugar recovery percentage of 99%. In this
tequila industry in Mexico and the Brazilian sugarcane stage, the major form of carbohydrate in agave (fructans) is
industry.26,27 The life-cycle stages for agave-derived ethanol converted to free sugars, such as fructose, which can later be
are agave cultivation, harvest and transport, and ethanol fermented to produce ethanol. The energy use for the diffuser
production. In this scenario, the sugar extracted from agave system and hydrated ethanol yield from sugarcane using the
juice is used for ethanol production, while the cellulosic diffuser system are adopted from Yan et al.26 In this ethanol
residues are combusted in a co-generation system to provide production scenario, enzymatic hydrolysis is adopted for agave
process energy with excess electricity exported to the electrical juice, although a short thermal treatment (using the waste heat
grid.26 We take into account the life-cycle fossil energy use and from the co-generation system) is used after enzymatic
GHG emissions for direct energy and material inputs, activities to accelerate the hydrolysis.26 Electricity is used for
machinery, and buildings. We account for CO2, CH4, and dehydration of ethanol and for stillage treatment. Stillage
N2O as GHG emissions, with all emissions converted to CO2 generation is assumed to be 12 L L−1 of ethanol26,27 (see pages
equivalents based on their 100 year global warming potential.26 S5−S13 of the Supporting Information).
We consider an intensive cultivation scenario with high levels The energy values and GHG offsets assigned to electricity
of fertilizer applications and soil amendments to reduce the export from agave are the life-cycle fossil fuel energy use and
harvest cycle to 6 years. Agaves are started from rhizomes, with GHG emissions to produce this electricity based on the U.S.
a planting density of 3350 plants per ha−1. 26 Following Yan et generation mix. Life-cycle fossil energy use and GHG emissions
al., the application rates of herbicides and pesticides are taken as for the U.S. generation mix are 2.91 MJ and 193 g of CO2 equiv
average values reported for corn and sugarcane and the fertilizer MJ−1 of electricity produced, respectively.26,29 The net GHG
application is 600, 40, and 240 kg ha−1 (for a 6 year cycle) as offsets for different yield scenarios of agave are calculated as the
nutrients of N, P, and K, respectively.26−28 In addition, ash fossil fuel energy saved by manufacturing ethanol from agave
from the combustion of bagasse and residue and filter cake compared to gasoline (94 g of CO2 equiv to produce 1 MJ of
from juice filtration during the biorefinery stage are also added gasoline) in addition to fossil fuel energy saved by exporting
as fertilizers.26,27 The mechanization level in the cultivation electricity by combustion of agave residues.19,26 In other words,
stage is assumed to be similar to the Brazilian sugarcane the total GHG offsets include offsets by agave-derived ethanol
industry27 (see pages S5−S13 of the Supporting Information). displacing gasoline and surplus electricity displacing grid
Harvesting is assumed to be mechanical using modified electricity minus the GHG emission resulting from the
whole-stick cane harvesters.26,27 Whole plants, including the production of agave-derived ethanol and electricity.
sugar-rich stems and leaves, are harvested. Yield (above-ground Solar PV Life-Cycle Analysis. We considered a solar PV
dry matter) is a major unknown variable in agave cultivation, installation, because PV is the dominant technology for current
and reported yield measurements range from 10 to 34 Mg ha−1 and proposed solar installations. Further, there might be other
year−1 for a precipitation range from 300 to 800 mm.15,16 logistic constraints for colocation of biofuels in CSP
However, higher yields are also reported around 42 Mg ha−1 installations because of intensive infrastructure.30 The solar
year−1.17,22,23 For comparison, the average yields reported for PV infrastructure is installed in a desert environment with an
corn and sugarcane are 10 Mg ha−1 year−1 (grain and stover) annual precipitation of 250 mm and a solar insolation of 2100
and 21 Mg ha−1 year−1 (sugar and bagasse), respectively.15 We kWh m−2 year−1. We adopt the design configuration of a very
consider three scenarios, low yield, baseline yield, and high large-scale PV power generation infrastructure described by Ito
yield, with annual yields of 10, 15, and 34 Mg ha−1 year−1, et al.31 This installation consists of a basic array of fixed flat
respectively. In the low-yield scenario, no irrigation is provided plate systems with approximately 3500 multi-crystalline silicon
and the only water inputs are from precipitation (assumed to be [m-Si modules of 120 Wp (watt peak, maximum power in watts
250 mm year−1). In the baseline- and high-yield scenarios, generated from a PV module under optimum conditions), with
3023 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030
Environmental Science & Technology Article

a module area of 1 m2] PV modules with an efficiency of 13%.31 operation costs were taken as $4 per Wp.36−38 Information
The performance ratio (PR, defined as the ratio of energy on the cost of cultivation and ethanol processing from agave is
output measured to energy output modeled, is a performance scarce. In this study, we used cultivation39 and ethanol
indicator of a solar infrastructure at a given location or processing39,40 costs of $1250, $1750, and $3250 ha−1 for
environmental conditions) of this PV infrastructure is assumed low-, baseline-, and high-yield scenarios of agave, respectively.39
to be 70%, which is typical of desert areas.32 The annual power The wholesale ethanol price used in this analysis was $2.75 per
generation is calculated as follows: gallon 41 (see pages S16 and S17 of the Supporting
Information).
annual power generation (kWh year −1) To illustrate the tradeoffs and synergies of colocating solar
= solar insolation (kWh m−2 year−1) × efficiency and biofuel infrastructures, we calculate the revenue generated
(U.S. $) per cubic meter of water used in the San Bernardino
× module area (m 2) × performance ratio County in California for agriculture, solar, and agave biofuel.
San Bernardino County was chosen as an example to illustrate
The materials and energy inputs and GHG emissions and the potential to colocate solar and biofuel projects because this
outputs during life cycles of m-Si modules are considered from
region has been identified as one of the major areas for large
PV production plants. The life-cycle stages considered are
manufacturing PV modules and balance of system (BOS) solar energy development. The region is characterized by an
components, construction and operation, decommission, and arid climate with around 250 mm of annual precipitation, and
recycling,33 assuming a life cycle of 30 years. Large-scale solar large areas are irrigated for crop and range production. We used
installations in deserts require balance of system components, the irrigated area and applied water estimates for various crops
such as module frames, mounting structures, grid connectors, and their values (U.S. $) in 2003 from the California
concrete, and office facilities.31,33 The energy inputs for Department of Water Resources and United States Department
producing 1 MWp [megawatt peak, maximum power (in of Agriculture (USDA), National Agricultural Statistics Service
MW) generated from a module under optimum conditions] of (NASS).42,43 The total value of all crops was divided by the
m-Si module and BOS components total 31 333 GJ.31 The total water applied to calculate the returns per amount of water
GHG emissions resulting from production of 1 kWh (kilowatt used.
hour) of m-Si module and BOS components are 37 and 20 g of Sensitivity and Uncertainty Analyses. Sensitivity anal-
CO2 equiv, respectively.33,34 We also consider the energy used ysis was performed for solar PV installation and agave-derived
for operation of the PV infrastructure, mainly for the routine ethanol. We defined a base case of the parameters considered,
cleaning of panels. In this study, we do not consider some identified a range of uncertainty for each parameter, and then
additional pollutants and heavy metal emissions, including tested the effect of changing each parameter (on energy output
direct and indirect emissions of cadmium.33 The energy values and GHG offsets) from its minimum to maximum value. We
and GHG offsets assigned to electricity produced by the solar used the module efficiency, insolation, performance ratio, and
PV infrastructure are the life-cycle fossil fuel energy use and
number of modules ha−1 for the solar PV infrastructure and
GHG emissions to produce this electricity based on the U.S.
generation mix (2.91 MJ and 193 g of CO2 equiv MJ−1 of overall sugar utilization efficiency and number of plants ha−1 for
electricity produced, respectively)26,29 (see page S14 of the the agave-derived ethanol system as input parameters (see
Supporting Information). pages S18−S21 of the Supporting Information).
To account for the dust or “soiling” impacts of solar power We addressed the uncertainty in our analysis using a Monte
production, a derate rate of 0.3% per day12 was applied to the Carlo simulation approach. The analysis was performed using
performance ratio for rainless periods, which extends over a the input values of the sensitive input parameters for solar PV
period of 7 months per year. The washing frequency was once and agave-derived ethanol, identified by sensitivity analysis. In
every week during the rainless periods (7 months) and once addition, we also included the parameters that affect the
every month for the rest of the year. The water use for per revenue and returns from solar infrastructure and agave
washing event was adopted from the planning reports for large cultivation. The input variables considered for the solar
PV solar power projects in the southwestern United States (20 installation were efficiency (11−15%), number of modules
m3 ha−1). Additional water application was used to suppress ha−1 (2500−3500), installation cost ($3−4 Wp−1), and the
dust production from the soil. We assume that the total annual wholesale price of solar electricity ($0.08−0.12 kWh−1). The
water requirement for dust management is approximately input variables considered for agave-derived ethanol were the
equivalent to 100 mm of annual precipitation. Water use for the overall sugar utilization efficiency (70−90%), number of plants
construction of PV infrastructure is also included in the life- ha−1 (2850−3850), cultivation and distillation cost ($1000−
cycle water use. Water requirements for washing panels and 3500 ha−1), and the wholesale price of biomass electricity
dust suppression are provided in page S15 of the Supporting
($0.08−0.12 kWh−1). The input variables were assumed to be
Information.
Integrated Solar Energy−Biofuel Systems. On the basis independent and were randomly selected from a uniform
of the life-cycle analyses, we evaluated the potential to integrate distribution, and the output simulation was repeated 104 times.
solar infrastructure with agave feedstock cultivation. We also The maximum, mean, minimum, and quantiles (5 and 95%) of
evaluated the economic returns [difference between annualized outputs for solar PV (outputs: energy input, energy output,
revenue and cost of installation (solar) or cultivation and GHG emissions, net GHG offsets, revenue, and returns) and
ethanol processing (agave)] from solar PV (electricity) and the three yield scenarios of agave (outputs: energy input, total
agave cultivation (ethanol and electricity export). The whole- energy output, ethanol energy, electricity export, GHG
sale electricity cost used in this analysis was $100 per emissions, net GHG offsets, revenue, and returns) were
MWh,35,36 and the lifetime (30 year) construction and reported (see pages S22−S25 of the Supporting Information).
3024 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030
Environmental Science & Technology Article

Figure 2. (a) Land area for solar infrastructures and ROW land area allotted for additional support facilities, such as transmission lines and roads.
The data include PV (n = 13), parabolic trough (n = 10), and central tower (n = 5) facilities. (b) Annual water requirements for large solar
infrastructures in the southwestern United States. The data include PV (n = 14), parabolic trough (n = 10), and central tower (n = 5) facilities. Error
bars represent the standard deviation for each solar technology.

■ RESULTS AND DISCUSSION

Analysis of the solar installations in the southwest U.S. indicates
that solar installations occupy significant land area and use a
considerable amount of water for construction and operation
(Figure 2). The land area required to install 1 MWp ranged
from 1.8 to 3.8 ha (Figure 2a). The ROW area (e.g., for access
roads and transmission lines) requested by solar companies is
typically twice the actual area of the installed solar infra-
structure (Figure 2a). The concentrated solar facilities with Figure 3. Partitioning of water use (average) in large solar
wet-cooling technology use high amounts of water compared to installations: PV and CSP (with dry cooling). The data include 5
PV systems (Figure 2b), resulting in 10 times more water use solar PV installations and 10 CSP installations (3 CSP parabolic
compared to dry-cooled CSPs. In addition to that, CSPs require trough with wet cooling, 3 CSP parabolic trough with dry cooling, and
some process water and water for routine mirror cleaning and 4 CSP central tower).
dust suppression (Figure 3). In the case of PV systems, water
use for construction and operation phases is mostly for dust dust suppression from soils by adding moisture, is a major
suppression and cleaning panels (Figure 3). Regardless of PV component of the total water use (Figure 3).
or CSP technology, water for dust management, either directly Sensitivity analysis indicated that the changes in the input
for washing dust from solar panels or indirectly for suppressing parameters, efficiency and number of modules ha−1 for solar PV
3025 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030
Environmental Science & Technology Article

Figure 4. Life-cycle energy inputs and outputs of solar PV and agave-derived ethanol. The error bars represent the uncertainty (maximum and
minimum) in our analysis determined by Monte Carlo analysis that varied the most important parameters, as determined by sensitivity analysis.

Figure 5. Life-cycle GHG emissions and net GHG offsets of agave-derive ethanol. The error bars represent the uncertainty (maximum and
minimum) in our analysis determined by Monte Carlo analysis that varied the most important parameters, as determined by sensitivity analysis.

and overall sugar utilization efficiency and number of plants yield), and from 29 to 39 Mg ha−1 (high yield). The 5 and 95%
ha−1 for agave, have significant impacts on the total energy quantile values are reported for all of the life-cycle model
output and GHG offsets (see more information on the analysis outputs. The maximum and minimum values of outputs are
on pages S18−S21 of the Supporting Information). The solar reported as error bars in Figures 4 and 5 and Table 1.
insolation and performance ratio are not expected to change The life-cycle analyses show that energy output from the
significantly as the panels are installed in a desert environment, solar PV system [energy output of 2405 GJ ha−1 year−1 (1940−
such as southwestern U.S. (minor variations in solar insolation) 2920)] is much larger than that from agave, with a mean energy
and are cleaned routinely to minimize the impacts as a result of output of 206 GJ ha−1 year−1 (171−243) for high yield, 87 GJ
soiling (minor variations in PR). We have also included ha−1 year−1 (73−103) for baseline yield, and 61 GJ ha−1 year−1
economic parameters in the uncertainty analysis: the (50−71) for low yield (Figure 4). The energy ratio (energy
installation cost of solar infrastructure and wholesale price of output to input) was around 5.4 (5−6) for solar PV installation.
solar electricity for solar PV and the cultivation and distillation For agave-derived ethanol, this ratio ranged from 3.3 (3.1−3.5),
costs and the wholesale price of fuel ethanol. Even though there 3.2 (3−3.6), and 4.8 (4.4−5.2) for the low-, baseline-, and high-
are several factors, which may affect the energy output and yield scenarios of agave. The energy ratios are consistent with
GHG offsets from solar PV and agave-derived ethanol, the existing life-cycle studies of large solar installations and agave-
factors that we considered are important in the case of derived ethanol production.26,31
colocation (see pages S22−S25 of the Supporting Information). The energy output from solar PV (mean) is 12, 28, and 40
The Monte Carlo approach indicated an agave yield range from times higher than that of high-, baseline-, and low-yield
8 to 12 Mg ha−1 (low yield), from 13 to 17 Mg ha−1 (baseline scenarios of agave. Despite the much higher rates of energy
3026 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030
Environmental Science & Technology Article

output for solar PV, the annual returns on investment ha−1

returnsb ($ ha−1 year−1)

10885 (−9859−32739)
(mean) are only 3, 9, and 14 times higher than that for high-,

3576 (2309−4965)
1146 (553−1769)

758 (316−1214)
baseline-, and low-yield scenarios of agave, respectively,
reflecting the higher value of the liquid fuel compared to
electricity (Table 1).
GHG emissions from the solar PV system [37 Mg of CO2
equiv ha−1 year−1 (45−30)] are much larger than those from
agave, with GHG emissions of 4.7 Mg of CO2 equiv ha−1 year−1
(4.2−5.1) for high yield, 2.9 Mg of CO2 equiv ha−1 year−1
revenue ($ ha−1 year−1)
66859 (49796−86583)

(2.6−3.2) for baseline yield, and 2.1 Mg of CO2 equiv ha−1

6828 (5593−8204)
2895 (2365−3466)

2008 (1642−2408)

year−1 (1.9−2.4) for low yield (Figure 5). The GHG emissions
ha−1 year−1 from solar PV (mean) are 8, 13, and 18 times
higher than those from the high-, baseline-, and low-yield
scenarios of agave, respectively (Figure 5). The GHG offsets
from the solar PV system [464 Mg of CO2 equiv ha−1 year−1
(375−562)] are much larger than those from agave, with GHG
offsets of 18 Mg of CO2 equiv ha−1 year−1 (14−22) for high
(Mg of CO2 equiv ha−1 year−1)

yield, 6.8 Mg of CO2 equiv ha−1 year−1 (5.3−8.2) for baseline

yield, and 4.6 Mg of CO2 equiv ha−1 year−1 (3.7−5.5) for low
GHG emissions

4.7 (4.2−5.1)
2.9 (2.6−3.2)

2.1 (1.9−2.4)
37 (45−30)

yield. GHG offsets (mean) resulting from electricity generation

by solar PV are substantially higher than those from agave-
derived ethanol, with offsets 26, 68, and 100 times higher than
The 5−95% quantile values are in parentheses. bCalculated by subtracting the cost of installation/cultivation from the revenue.

high-, baseline-, and low-yield scenarios of agave (Table 1).

To compare the land use efficiency for different biofuel
feedstocks, we compared the life-cycle energy and net GHC
Table 1. Summary of Annual Life-Cycle Energy, GHG, Water, and Revenues from Agave and Solar PVa

emissions from agave-derived ethanol to ethanol derived from

sugarcane, corn grain, and switchgrass from existing studies.26
On the basis of the results from Yan et al., the land use
(GJ ha−1 year−1)
449 (394−506)
energy input

efficiency (energy output and GHG emissions) of agave can be

43 (39−47)
27 (24−29)

18 (16−20)

higher or lower than that of sugarcane, depending upon the

yield scenario considered, while agave is comparable to corn
and switchgrass even when we consider the low-yield
scenario.26 A particularly important issue is the considerable
uncertainty in yield estimates for agave, with most of the
(Mg of CO2 equiv ha−1 year−1)

available estimates being based on extrapolation from measure-

ments on single plants.16 Reliable yield estimates of various
agave cultivars are critical for establishing the long-term
464 (375−562)

6.8 (5.3−8.2)

4.6 (3.7−5.5)
GHG offsets

18 (14−22)

economic and logistic viability of commercial agave-derived

ethanol production. To this end, long-term studies over large
areas are necessary to obtain realistic estimates of yields and
understand the ecophysiological responses of agave interspaced
with solar infrastructure.
A life-cycle analysis of a hypothetical colocation of solar PV
and agave for San Bernardino County in California indicated
higher returns per m3 of water used than either system alone
2405 (1940−2920)

(Table 2). The solar PV and agave both can achieve high
(GJ ha1 year−1)

206 (171−243)
energy output

87 (73−103)

61 (50−71)

revenues per m3 of water used compared to traditional

agricultural crops. The revenues from solar PV per unit of
water used were over 100 times higher than traditional crops,
while the revenues from baseline- and high-yield scenarios of
agave were 5 and 2 times higher, respectively. The lower
revenue per unit of water consumed for the high-yield scenario
(m3 ha−1 year−1)

of agave can be attributed to high irrigation water input.

water

1050
5500
1000

However, we have to consider that installation costs are very

high for solar infrastructures. Total returns on investment for
solar PV, baseline-yield agave, and high-yield agave were 10,
1.3, and 1 times higher than the revenue of agricultural crops.
Returns per m3 water use of these energy systems are higher
agave, high yield

agave, low yield

agave, baseline

than even the revenue per m3 water use of agricultural crops.

land use

The combination of solar PV and baseline-yield scenario of

solar PV

yield

agave achieved the highest returns per m3 of water used,

highlighting the synergies of colocation.
a

3027 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030

Environmental Science & Technology Article

Table 2. Ratio of Revenue (and Returns) to Water Use for Agave and Solar PV as Hypothetical Independent and Colocated
Systems in San Bernardino County, CAa
independent ($ m−3 ha−1 year−1) colocated ($ m−3 ha−1 year−1)
land use revenue returns revenue returns
crops 0.6
solar PV 64 (48−82) 10 (−9−31)
agave, baseline yield 3 (2.4−3.5) 1 (0.6−1.8) 66 (50−86) 11 (−8.9−32)
agave, high yield 1.2 (1−1.5) 0.6 (0.4−0.9) 12 (9.2−16) 2.4 (−1.2−6.3)
a
The 5−95% quantile values are in parentheses.

In our analysis, the synergy between PV and agave stems modules ha−1) in colocated systems, to account for the space
from the fact that water addition for washing panels is beneficial requirements for plants and routine agricultural operations.
for both the solar PV electricity output and agave yield. For Another consideration is the possible shading of solar modules
example, we show that dust deposition can be a major factor by plants in colocated systems. In the case of agave, many
affecting the economics of large solar infrastructures (annual cultivated species reach only a size of 1.5−2 m in 5−7 years and
loss of 15%) and, with proper dust control measures (routine produce long flowering stalks (2−10 m) only after 6 years. In
cleaning operations), the losses can be reduced to under 1%. colocated systems, harvesting is expected to occur after 6 years
Other potential synergistic factors of integrated solar−agave when the agave plants are not tall enough to shade the panels.
systems include enhanced soil moisture availability to agave as a The intensive agricultural operations, such as planting, fertilizer
result of the concentration of rainfall (or water used for application, and use of whole-stick cane harvesters, may have an
washing panels) by solar panels into the interspaces, reduced impact on background dust emissions from these systems. In
soil erosion by wind and water, and decreased dust emissions this study, we used the upper end values of water utilization for
from disturbed soils as a result of moisture addition and cleaning solar installations and dust suppression to account for
vegetation (agave) cover. Moreover, many Agave spp. are water requirements for these additional impacts. The feasibility
known to benefit from an increase in the CO2 concentration in of colocation also depends upon availability of ethanol-refining
the atmosphere and tolerate high temperatures (in the range of facilities near the sites selected for colocation. Moreover,
60 °C) and recurrent droughts,22,44 which are expected to be considerable uncertainties remain on the agave yield and the
more frequent in the future.45 benefits of periodic water addition (light irrigation) and shading
Overall, colocated solar and agricultural infrastructure could from the solar infrastructures in deserts. Colocation may not be
(1) ensure efficient water use by sustaining biofuel production practical in all locations, and further field studies are required to
and maximizing power output form solar installations in fully evaluate the advantages and disadvantages of colocation.
deserts, (2) minimize dust generation, and (3) minimize In the southwestern United States, solar deployments are
impacts on natural areas by deploying biofuel cultivation in increasing rapidly, with growing concerns about the fate of soil
existing large solar facilities. Our results indicate that solar PV and water resources in these areas.46−48 The western United
facilities use much less water compared to other energy States accounted for half of all U.S. population growth in the
technologies46 and traditional crops.43 In our analysis, the past decade, creating additional demand for land and water
annual water requirement for energy production for the solar resources.47,48 Moreover, the southwest has experienced rapid
infrastructure was 0.43 L MJ−1 (0.54−0.36) compared to 11.5 L warming and recurrent droughts,45,49 further increasing the
MJ−1 (13.7−9.5) and 26.7 L MJ−1 (32.9−22.6) for baseline- pressure on soil and water resources. In these water-limited
and high-yield scenarios of agave, respectively. A life-cycle areas, coupled solar infrastructure and biofuel cultivation could
analysis of a hypothetical colocation indicated higher returns be established in marginal lands with low water use, thus
per m3 of water used than either system alone. Water minimizing the socioeconomic and environmental issues
requirements to produce energy were 0.22 L MJ−1 (0.28− resulting from cultivation of biofuel crops in prime agricultural
0.19) and 0.42 L MJ−1 (0.52−0.35) for solar PV−agave lands.
(baseline yield) and solar PV−agave (high yield), respectively. The United States Department of Energy estimates that
However, the water requirement for agave is low compared to around 850 000 ha of direct land transformation would be
agricultural and other biofuel crops, and agave can survive and required to install the 2030 scenario of 302 GW of solar PV.36
produce significant biomass in nutrient-poor soils and adverse Assuming that all of these areas could be integrated with agave
climatic conditions, thanks to their water-efficient CAM ethanol production, around 1−2 billion gallons of fuel ethanol
photosynthesis pathway.21,22 In particular, the baseline yield (and additional electricity exports) could be produced.
scenario for agave requires no additional water inputs other Although this is a small percentage (3−6%) of the United
than the water required for operation of the solar infrastructure. States 2030 ethanol production target of 30 billion gallons,50 we
In this scenario, even though energy outputs from agave- have to consider that this ethanol is produced from marginal
derived ethanol are much lower compared to the solar PV, they lands with no additional land transformation.

■
provide a more valuable (and easily transportable) form of
energy in liquid fuels. Thus, in some dryland regions, colocated ASSOCIATED CONTENT
solar PV−agave systems may provide attractive economic
incentives in addition to efficient land and water use. *
S Supporting Information

In some existing solar facilities, it might be possible to Additional information regarding our approach and analysis
integrate agave cultivation. Design modifications could be (additional data, methods, assumptions, figures, tables, and
considered in future PV to better integrate agave cultivation. A references). This material is available free of charge via the
major consideration is the spacing of PV panels (and number of Internet at http://pubs.acs.org.
3028 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030
Environmental Science & Technology

■
Article

AUTHOR INFORMATION (18) Tilman, D.; Socolow, R.; Foley, J. A.; Hill, J.; Larson, E.; Lynd,
L.; Pacala, S.; Reilly, J.; Searchinger, T.; Somerville, C.; Williams, R.
Corresponding Author Beneficial biofuelsThe food, energy, and environment trilemma.
*Telephone: 00-1703-581-8186. Fax: 650-498-5099. E-mail: Science 2009, 325, 270−271.
[email protected]. (19) Farrell, A. E.; Plevin, R. J.; Turner, B. T.; Jones, A. D.; O’Hare,
Notes M.; Kammen, D. M. Ethanol can contribute to energy and
environmental goals. Science 2006, 311, 506−508.
The authors declare no competing financial interest.

■
(20) El Bassam, N. C3 and C4 plant species as energy sources and
their potential impact on environment and climate. Renewable Energy
ACKNOWLEDGMENTS 1998, 15, 205−210.
The authors acknowledge the TomKat Center for Sustainable (21) Borland, A. M.; Griffiths, H.; Hartwell, J.; Smith, J. A. C.
Exploiting the potential of plants with crassulacean acid metabolism
Energy at Stanford University. Symbols for Figure 1 are
for bioenergy production on marginal lands. J. Exp. Bot. 2009, 60,
courtesy of the Integration and Application Network (ian. 2879−2896.
umces.edu/symbols). (22) Noble, P. S. Environmental Biology of Agaves and Cacti;

■ REFERENCES
(1) Intergovernmental Panel on Climate Change (IPCC). Special
Cambridge University Press: Cambridge, U.K., 1988.
(23) Nobel, P.; Garciamoya, E.; Quero, E. High annual productivity
of certain agaves and cacti under cultivation. Plant Cell Environ. 1992,
15, 329−335.
Report on Renewable Energy Sources and Climate Change Mitigation;
Cambridge University Press: New York, 2011. (24) California Energy Commission; http://www.energy.ca.gov/.
(2) Youngs, H.; Somerville, C. Development of feedstocks for (25) Bureau of Land Management; http://www.blm.gov/.
cellulosic biofuels. F1000 Biol. Rep. 2012, 4, 10. (26) Yan, X.; Tan, D. K. Y.; Inderwildi, O. R.; Smith, J. A. C.; King,
(3) Cabraal, R. A.; Barnes, D. F.; Agarwal, S. G. Productive uses of D. A. Life cycle energy and greenhouse gas analysis for agave-derived
energy for rural development. Annu. Rev. Environ. Resour. 2005, 30, bioethanol. Energy Environ. Sci. 2011, 4, 3110−3121.
117−144. (27) Macedo, C.; Seabra, J. E. A.; Silva, J. A. E. R. Green house gases
(4) Burney, J. A.; Woltering, L.; Burke, M.; Naylor, R.; Pasternak, D. emissions in the production and use of the ethanol from sugarcane in
Solar-powered drip irrigation enhances food security in the Sudano− Brazil: The 2005/2006 averages and a prediction for 2020. Biomass
Sahel. Proc. Natl. Acad. Sci. U. S. A. 2010, 107 (5), 1848−1853. Bioenergy 2008, 32, 582−595.
(5) Goldemberg, J. Ethanol for a sustainable energy future. Science (28) Renewable and Applicable Energy Laboratory (RAEL). Energy
2007, 315, 808−810. and Resources Group (ERG) Biofuel Analysis Meta-Model (EBAMM);
(6) Committee on Water Implications of Biofuels Production in the University of California: Berkeley, CA, 2007.
United States, Water Science and Technology Board (WSTB), (29) Kim, S.; Dale, B. E. Life cycle inventory information of the
Division on Earth and Life Studies (DELS), and National Research United States electricity system. Int. J. Life Cycle Assess. 2004, 10, 294−
Council. Water Implications for Biofuel Production in the United States; 304.
The National Academies Press: Washington, D.C., 2008. (30) Burkhardt, J.; Heath, G.; Turchi, C. Life cycle assessment of a
(7) Pimentel, D.; Marklein, A.; Toth, M. A.; Karpoff, M.; Paul, G. S.; parabolic trough concentrating solar power plant and impacts of key
McCormack, R.; Kyriazis, J.; Krueger, T. Biofuel impacts on world design alternatives. Environ. Sci. Technol. 2011, 45, 2457−2464.
food supply: Use of fossil fuel, land and water resources. Energies 2008, (31) Ito, M.; Kato, K.; Komoto, K.; Kichimi, T.; Kurokawa, K. A
1, 41−78. comparative study on cost and life-cycle analysis for 100 MW very
(8) Tsoutsos, T.; Frantzeskaki, N.; Gekas, V. Environmental impacts large-scale (VLS-PV) systems in deserts using m-Si, a-Si, CdTe, and
from solar energy technologies. Energy Policy 2005, 33, 289−296. CIS modules. Prog. Photovoltaics 2008, 16, 17−30.
(9) United Nations Energy Statistics; http://unstats.un.org/unsd/ (32) Kawajiri, K.; Oozeki, T.; Genchi, Y. Effect of temperature on PV
energy/. potential in the world. Environ. Sci. Technol. 2011, 45, 9030−9035.
(10) Millennium Ecosystem Assessment (MEA). Ecosystems and (33) Fthenakis, V. M.; Kim, H. C. Photovoltaics: Life-cycle analyses.
Human Well Being: Desertification Synthesis; World Resource Institute: Sol. Energy 2011, 85, 1609−1628.
Washington, D.C., 2005. (34) Raugei, M.; Bargigli, S.; Ulgiati, S. Life cycle assessment and
(11) Karekezi, S.; Kithyoma, W. Renewable energy strategies for rural energy pay-back time of advanced photovoltaic modules: CdTe and
Africa: Is a PV-led renewable energy strategy the right approach for CIS compared to poly-Si. Energy 2007, 32, 1310−1318.
providing modern energy to the rural poor of sub-Saharan Africa? (35) Solar Cell Central. Solar Electricity Costs; http://solarcellcentral.
Energy Policy 2002, 30, 1071−1086. com/cost_page.html.
(12) Kimber, A.; Mitchell, L.; Nogradi, S.; Wenger, H. The effect of (36) U.S. Department of Energy. SunShot Vision Study; U.S.
soiling on large grid-connected photovoltaic systems in California and Department of Energy: Washington, D.C., 2012; DOE/GO-102012-
the southwest region of the United States. Conference Record of the 3037, http://www1.eere.energy.gov/solar/pdfs/47927.pdf.
2006 IEEE 4th World Conference on Photovoltaic Energy Conversion (37) Open Energy Information; http://en.openei.org.
2006, 2, 2391−2395; http://ieeexplore.ieee.org/stamp/stamp.jsp?tp= (38) Goodrich, A.; James, T.; Woodhouse, M. Residential,
&arnumber=4060159. Commercial, and Utility Scale Photovoltaic (PV) Systems in the United
(13) Ibrahim, M.; Zinßer, B.; El-Sherif, H.; Hamouda, E.; Makrides, Statees: Current Drivers and Cost Reduction Opportunities; National
G.; Georghiou, G. E.; Schubert, M. B.; Werner, J. H. Symposium Renewable Energy Laboratory (NREL): Golden, CO, 2012; p 64,
Photovoltaische Solarenergie; Bad Staffelstein, Germany, March 4−6, NREL TP-6A20-53347.
2009; pp 318−323. (39) Nunez, H.; Rodriguez, L. F.; Khanna, M. Agave for tequila and
(14) Mani, M.; Pillai, R. Impact of dust on solar photovoltaic (PV) biofuels: An economic assessment and potential opportunities. GCB
performance: Research status, challenges and recommendations. Bioenergy 2011, 3, 43−57.
Renewable Sustainable Energy Rev. 2010, 14, 3124−3131. (40) Crago, C. L.; Khanna, M.; Barton, J.; Giuliani, E.; Amaral, W.
(15) Somerville, C.; Youngs, H.; Taylor, C.; Davis, S. C.; Long, S. P. Cost competitiveness of sugarcane ethanol in Brazil relative to corn
Feedstocks for lignocellulosic biofuels. Science 2010, 329, 790−792. ethanol in the US. Energy Policy 2010, 38, 7404−7415.
(16) Davis, S. C.; Dohleman, F. G.; Long, S. P. The global potential (41) United States Department of Agriculture (USDA). Livestock and
for agave as a biofuel feedstock. GCB Bioenergy 2011, 3, 68−78. Grain Market News, National Weekly Ethanol Summary; USDA:
(17) Garcia-Moya, E.; Romero-Manzanares, A.; Nobel, P. S. Washington, D.C., 2012; http://www.ams.usda.gov/mnreports/
Highlights for agave productivity. GCB Bioenergy 2011, 3, 4−14. lswethanol.pdf.

3029 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030

Environmental Science & Technology Article

(42) California Department of Water Resources; http://wwwdwr.

water.ca.gov/.
(43) United States Department of Agriculture (USDA), National
Agriculture Statistic Service (NASS); http://www.nass.usda.gov/.
(44) Drennan, P. M.; Noble, P. Responses of CAM species to
increasing atmospheric CO2 concentrations. Plant Cell Environ. 2000,
23, 767−781.
(45) Seager, R.; Ting, M.; Held, I.; Kushnir, Y.; Lu, J.; Vecchi, G.; H.
P. Huang, H. P.; Harnik, N.; Leetmaa, A.; Lau, N.-C.; Li, C.; Velez, J.;
Naik, N. Model predictions of an imminent transition to a more arid
climate in southwestern North America. Science 2007, 316, 1181−
1184.
(46) Fthenakis, V.; Kim, H. C. Life-cycle uses of water in U.S.
electricity generation. Renewable Sustainable Energy Rev. 2010, 14,
2039−2048.
(47) Carter, N. T.; Campbell, R. J. Water Issues of Concentrating Solar
Power (CSP) in the U.S. Southwest; Congressional Research Service:
Washington, D.C., 2009; Order R40631.
(48) Anderson, M. T.; Woosley, L. H. Water Availability for the
Western United StatesKey Scientific Challenges; United States
Geological Survey (USGS): Reston, VA, 2005; U.S. Geological Survey
Circular 1261.
(49) U.S. Global Change Research Program. Global Climate Change
Impacts in the United States; Cambridge University Press: New York,
2009.
(50) The Capitol.Net, Inc. Energy: Ethanol, United States Government
Series; The Capitol.Net, Inc.: Alexandria, VA, 2010.

3030 dx.doi.org/10.1021/es404950n | Environ. Sci. Technol. 2014, 48, 3021−3030